Meteorites rain down on the whole planet more or less evenly. The record of them does not. Of the 45,715 meteorites in the Meteoritical Society's catalogue as NASA publishes it, 65% were picked up on Antarctic ice and 24% in hot deserts. The 1,107 that someone actually saw fall tell a different story again: they follow people.
Every day, rocks from space reach the ground somewhere on Earth. Most fall into the sea, onto forests or fields, and are never seen again. The ones we know about were either seen to fall, and then picked up (a "fall"), or simply spotted on the ground later (a "find"). The Meteoritical Society has kept the official list for decades, and NASA publishes a copy with the place, year, class and weight of each one.
Put that list on a map and it looks like a map of meteorites. It is really a map of searching. Antarctica, where ice flows carry stones to bare blue ice fields and expose them, and the hot deserts of North Africa and Oman, where dark stones sit for thousands of years on pale ground, account for 89% of all records. The witnessed falls, which do not depend on searching, crowd into Europe, India and Japan, where there have long been many people to look up and someone to write it down. Use the buttons to switch between falls and finds, filter by year and class, and click any dot.
Dots: the 32,186 meteorites (70%) with a recorded place, sized by recovered weight when "Size by weight" is on. Witnessed falls are drawn on top with a dark ring. 13,529 records have no usable place (most are desert stones bought from dealers and named only for a broad area, such as 4,300 of the 4,499 called "Northwest Africa"); they count in the totals but cannot be mapped. Density counts the dots that pass the filters in each small square of the screen. Source: The Meteoritical Society, Meteoritical Bulletin Database, via NASA's Open Data Portal. Outlines: Natural Earth. Base map tiles © Esri.
Only 1,107 of the 45,715 meteorites (2.4%) were seen to fall. The other 44,608 are finds, and most come from a handful of places. Antarctica alone has 29,783 (65%), the Sahara 6,754 (15%), Arabia, mostly Oman, 3,163 (7%), and the Atacama and Australian deserts 905. The whole rest of the world has 3,571 (8%), and 3% cannot be placed at all.
Where the record comes from
Share of meteorites by where they were recovered: all records, finds, and witnessed falls. Hover a block.
Turn it around and the witnessed falls are almost all somewhere else: 96% of them land in the "rest of the world", and not one has been recorded in Antarctica. If falls were recorded evenly across the land, Antarctica, which is 8% of it, would have about 92 of them.
The biggest collection areas
Meteorites are named after the place they were found, so the name tells you the area. The ten largest hold 63% of the whole record. Click a row to zoom there.
Before 1969 the record grew slowly, by chance finds and witnessed falls. The first Antarctic meteorite, Adelie Land, turned up in 1912, and only 4 more followed before 1969, when a Japanese team found 9 near the Yamato Mountains and realised the ice was concentrating them. Systematic searches by Japanese and American teams followed, and the name Yamato alone now covers 7,269 meteorites. From the 1980s hunters and dealers began combing the Sahara and Oman. Finds peaked in the 2000s, with 17,697 in one decade; 97% of all dated finds are from 1970 or later.
Meteorites recorded per decade
Finds by where they were recovered (bars) and witnessed falls (line, right scale), 1800s to 2010s. The record ends in 2013, so the 2010s are partial. Hover a bar.
Antarctic finds are also small. The median Antarctic meteorite weighs 13.4 g, the median Saharan one 235 g, and the median find from the rest of the world 1,700 g: on bare ice or sand, a stone the size of a grape is easy to spot, while in a field it is just another pebble. Many of these stones are also pieces of the same fall, scattered across the ice and named separately until someone pairs them, so the counts overstate the number of separate falls.
A map of meteorites is really a map of where people have looked for them.
Witnessed falls are the fairest test, because nobody goes out looking for them: they are seen, heard and then searched for. If every fall on land were noticed, each country's share of falls would match its share of land. Instead, Europe, with 16% of the land, has 34% of the falls (2.2 times its share); Asia, home to India's 125 falls, has 1.4 times; North America almost exactly its share; Africa 0.6, South America 0.4 and Oceania 0.3.
Witnessed falls by continent, against an even spread
Bars: falls recorded. Ticks: falls expected if the 1,094 falls with a country were spread evenly over the land, Antarctica included. Hover a bar.
Country by country the gap is wider. Japan has 42 falls where an even spread predicts 2.8, about 15 times as many; Germany, France and Italy 12 to 14 times; the United Kingdom 10 times, with 19 falls. India has 125, 5.3 times its share. At the other end Canada has 16 where 74 would be expected, Brazil 22 against 63, and Australia 15 against 57. Libya, which has 1,466 finds, has 1 fall.
Falls recorded compared with an even spread
Observed falls divided by expected falls. Top: countries with at least 1.5 falls expected. Bottom: the shortfalls among countries with at least 10 expected. 1 = exactly its share of the land.
The link with people is clear but loose. Across the 154 countries larger than 20,000 km², falls per square kilometre rise with people per square kilometre (Spearman's rank correlation 0.34); among the 106 that have at least one fall, the correlation of the two on a log scale is 0.62. Population here is today's rough estimate from Natural Earth, not the population when each stone fell, and history matters as much as headcount: Europe and Japan have centuries of written records, newspapers and museums that wanted the stones.
Falls and people, country by country
Each dot is a country larger than 20,000 km² with at least one fall: people per km² (Natural Earth estimate) against falls per million km². Both scales are logarithmic. Hover or tap a dot.
Over time, witnessed falls rose from 17 in the 1800s to 56 in the 1860s and a peak of 91 in the 1930s, as populations, literacy, newspapers and scientific interest grew. Then they levelled off, at about 59 a decade since 1950, even though the world's population grew by 98% between the 1960s and the 2000s. The oldest falls in the record are Nogata in Japan (dated 860) and Narni in Italy (920).
The heaviest meteorite on record is Hoba in Namibia, 60.0 tonnes, still lying where it was found in 1920. Next come Cape York in Greenland (58.2 t, many pieces) and Campo del Cielo in Argentina (50.0 t). All ten of the heaviest are irons, and 9 of the ten are finds. The heaviest witnessed fall is Sikhote-Alin in the Russian Far East (23 t, 1947), followed by Jilin in China (4 t, 1976) and Allende in Mexico (2 t, 1969).
The heaviest meteorites
Total recovered weight. Click a row to see it on the map.
Iron meteorites are rare among falls (4.4%) but make up 27% of finds outside Antarctica and the deserts. They rust slowly, look and feel unlike any local rock, and set off metal detectors, so in fields and forests they are the ones people notice and keep. On ice and sand, where every dark stone stands out, the mix of finds looks much more like the falls: mostly ordinary chondrites. The rarest stones of all, 119 from Mars and 165 from the Moon, come 95% from ice and desert.
What kind of meteorite, by how it was recovered
Share of each broad class among witnessed falls, Antarctic finds, hot desert finds and finds from the rest of the world. Hover a block.
Data. NASA's Meteorite Landings table (45,716 records), compiled from the Meteoritical Society's Meteoritical Bulletin Database, from a copy mirrored on GitHub and checked cell by cell against a second, independent mirror (identical except that three old finds with no year in the first copy have a garbled year in the second) and a third, five-column one. It has the name, class, recovered weight, fall or find, year and coordinates of each meteorite. The table stops in 2013; the live Bulletin has grown a great deal since. I dropped one record, Meridiani Planum, which was found on Mars by a rover. Years are as given (the table stores 1 January of the year); one year in the future was set to missing.
Places. 13,529 records have no coordinates or sit at exactly 0°, 0°, which means unknown. For these I took the region from the collection area in the name (for example "Northwest Africa 869" or "Yamato 791197") when at least 80% of that area's located meteorites share one region (11,990 records). Antarctica is everything south of 60°S. The hot deserts are rough boxes: the Sahara (15°N to 34°N, 17.5°W to 34°E), Arabia (12°N to 31°N, 35°E to 60°E), the Atacama, and the Australian deserts (including the Nullarbor). Present-day countries are from Natural Earth.
Even spread. Meteorites arrive at a rate that is close to uniform over the globe, so I compare each country's falls with its share of the world's land area. Population is Natural Earth's recent estimate (POP_EST), used only as a rough guide; it is not the population at the time of each fall. About 2.4 times as many falls land in the sea as on land, and none of those are in the record.
What this is not. It is not a count of meteorite falls. Antarctic and desert totals are inflated by fragments of single falls named separately; finds have no date of fall at all, only a date of finding; and weights are the total recovered, sometimes from many pieces. The record also reflects who was searching: national Antarctic programmes, commercial hunters in the Sahara and Oman, and the museums that classified their stones.